Browse the docs

Skills — Craft and build

/feature-design

Design it right the first time.

The core loopuser-invoked

Structured design thinking for a new feature. /feature-design takes a problem statement and produces the full shape of a solution: a stepped user flow with rationale, component decisions benchmarked against world-class products, a CRISP compliance check per step, and the open questions that need answers before anyone builds.

It is not wireframe generation. Every decision in the output is justified against a CRISP dimension, and any step that fails one is flagged with a proposed fix before the design moves on.

Reach for it at the start of feature work, before code:

  • A PM brief or user problem needs converting into a designed flow
  • You are about to replace a manual workaround with a real feature
  • An agent is about to build a feature and you want the design decided first, not improvised mid-build
  • You have /crisp-research output and want it turned into a grounded design
  • User flow with decision rationale
  • CRISP compliance check per step
  • Component recommendations
  • Research questions to validate
Claude Code — /feature-design
User triggers approval flow
Can approve inline? Check context
Yes → Inline card
✓ Seamless
No → Contextual redirect
⚠ Avoid if possible
Optimistic confirm → sync silently

Illustrative example output — not a real audit.

Problem framing comes first. If these are not answered, the command asks before designing anything:

  • Who is experiencing this problem — role, sophistication, context
  • What they are trying to do — the job, not the feature
  • What they do today — the workaround being replaced
  • What success looks like — an observable outcome, not feature completion
  • What constraints exist — technical, timeline, design system
Does it produce wireframes or mockups?

No. It produces the thinking a wireframe should be based on: flows, states, component choices, edge cases, and rationale. Pair it with /handoff to turn the approved design into an implementation-ready spec.

What if I only have a vague idea?

Run /crisp-brief first. It converts a vague request into a scoped .brief.md, which /feature-design then reads alongside .crisp.md. Starting from a sharp brief is the difference between designing a solution and decorating an assumption.

How does it use /crisp-research output?

Directly. The research skill's patterns, benchmarks, and open questions are structured to paste into /feature-design as context, so the design starts from evidence about how world-class products solve the problem.

What does the CRISP compliance summary add?

Every step of the flow is checked against the five dimensions, with risks named. It catches the classic failure of feature design — a flow that works on the happy path but fails Contextual in the empty state or Responsive under load — before it is built.

zsh
$ npx skills add @laith-wallace/crisp

Installs all fourteen CRISP skills and auto-detects your AI harness. Run /crisp-teach once per project first.

The CRISP Letter

Design evaluation, in writing.

Occasional letters on making AI agents produce work worth shipping. New skills announced here first. No noise.